I agree on #2 though, it doesn't make as much sense for Apple. This would sort of be diametrically opposed to the quite successful decision to use Intel CPUs.
However Apple's potential here is hugely overstated. ARM licensees aren't at the whim of ARM -- they knew to sign generous IP agreements that essentially allow them to fork the technology. You have Apple on one side, with essentially a unchanged A8 created by Samsung, and virtually every other tech company on the other side -- Apple would quickly find themselves in another PowerPC situation if they bought ARM.
Speaking of PowerPC, Intel has been showing off some very impression demos of a smartphone powered by an Atom. Android of course is built from the ground up to be generally processor agnostic, so it wouldn't be a huge migration.
Do you mean companies like Marvell have comprehensive licenses allowing them to design their ARM cpu at their own will?
1) No matter how big the Apple is, the market would create new CPU architectures that would eat Apple's initial monopoly advantage over time. Investments are made for long term gains, so in the long run its not a good investment.
2) Apple can already distinguish itself from competition by its engineering, philosophy etc. a cpu architecture is not going to make enough difference. It would be wiser to spend money on innovation than some cpu architecture that already exists and is beneficial to Apple.
You can probably compare this to HP buying Palm for WebOS. HP bought it for having a strong software portfolio for its mobile devices.
The difference though is that ARM has to rely on its ecosystem (e.g. all other partners who bundle their products with ARM cpus) to survive. It needs to be licensed to the market. In case of WebOS, it was only used by Palm itself.